CN107301489A - Implement the production system of the production schedule - Google Patents
Implement the production system of the production schedule Download PDFInfo
- Publication number
- CN107301489A CN107301489A CN201710243910.6A CN201710243910A CN107301489A CN 107301489 A CN107301489 A CN 107301489A CN 201710243910 A CN201710243910 A CN 201710243910A CN 107301489 A CN107301489 A CN 107301489A
- Authority
- CN
- China
- Prior art keywords
- unit
- product
- production system
- multiple species
- monitoring unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/18—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form
- G05B19/4155—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form characterised by programme execution, i.e. part programme or machine function execution, e.g. selection of a programme
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06313—Resource planning in a project environment
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/18—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form
- G05B19/406—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form characterised by monitoring or safety
- G05B19/4065—Monitoring tool breakage, life or condition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/31—From computer integrated manufacturing till monitoring
- G05B2219/31427—Production, CAPM computer aided production management
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/37—Measurements
- G05B2219/37229—Test quality tool by measuring time needed for machining
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Computer Interaction (AREA)
- Manufacturing & Machinery (AREA)
- Automation & Control Theory (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Biodiversity & Conservation Biology (AREA)
- Educational Administration (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- General Factory Administration (AREA)
Abstract
The present invention provides a kind of production system for implementing the production schedule.Rapid detection machinery there occurs failure etc. and effectively be acted.The unit control apparatus of production system includes：Product information monitoring unit, it monitors product information；Part supply condition monitoring unit, it monitors the quantity of supply various parts into the part and multiple species of multiple species of at least one unit；And notification unit, in the case that the quantity of its various parts in the multiple species monitored by part supply condition monitoring unit is not according to the preset range that various parts are determined in multiple species, notified to upper Management Controller.
Description
Technical field
The present invention relates to a kind of production system for the production schedule for implementing to map out in upper Management Controller.
Background technology
In the past, on assembly line etc., in order to manufacture product, operate multiple parts of multiple species.Fig. 8 is in the prior art
Production system block diagram.In fig. 8, unit 400 includes multiple mechanical R1~R2 and control machinery R1~R2 multiple machines
Tool control device RC1~RC2.It is set in unit 400 multiple mechanical R1~R2 individually or manufactures product with cooperating.And
And, communicably it is connected as the upper Management Controller 200 of production schedule device by communication unit 410 with unit 400.
Here, the species of part and the quantity of all parts required for a product will be manufactured as product information S0,
It is contained in upper Management Controller 200.Upper Management Controller 200 supplies what is determined according to product information S0 to unit 400
The part of multiple species of quantity.
In addition, disclosing portion of the production schedule device based on multiple species in Japanese Unexamined Patent Publication 2013-016087 publications
The stock of part or the quantity information of part improve productive technology.
The content of the invention
However, in the production system shown in Fig. 8, when mechanical R1~R2 of unit 400 there occurs failure, or operator
, sometimes will not be according to the production meter set by upper Management Controller 200 in the case of the upper Management Controller 200 of maloperation
Draw and make product.
Specifically, there is following situation.
(1) because at least one in mechanical R1~R2 there occurs failure, therefore manufacturing capacity is remarkably decreased.
(2) although the supervision scope of upper Management Controller 200 is wider, response is low.Therefore, the supply of part
It is relatively slow, so that machinery R1~R2 turns into the state for waiting part, generation time loss.Now, it is impossible to suitably manufacture product.
(3) input to upper Management Controller 200 is wrong, and part supply produces superfluous and not enough.
In the case where not detecting these problem points rapidly, pass through over time, the manufacture efficiency of production system declines.
In addition, in Japanese Unexamined Patent Publication 2013-016087 publications, the technology of undisclosed rapid detection above-described problem.
The present invention is to complete in light of this situation, and its object is to can detect mechanical hair rapidly there is provided one kind
Failure etc. and the production system effectively acted are given birth to.
To achieve these goals, possessed according to the 1st invention there is provided a kind of production system：At least one unit, it is wrapped
The multiple machine control units for including multiple machineries of manufacture product and being controlled to the plurality of machinery；Unit control apparatus, its
Communicably it is connected to control the unit with least one unit；And upper Management Controller, its communicably with this
Unit control apparatus is connected, and including product information, the product information includes being used for multiple kinds that manufacture a product
The quantity of various parts in the part of class and the multiple species, the unit control apparatus possesses：Product information monitoring unit, its
Monitor the product information；Part supply condition monitoring unit, it monitors supply to the multiple kind of at least one unit
The quantity of various parts in the part of class and the multiple species；And notification unit, it is by the part supply condition monitoring unit
The quantity of various parts is not in making a reservation for for being determined according to various parts in the multiple species in the multiple species monitored
In the range of in the case of, notified to the upper Management Controller.
According to the 2nd invention, in the 1st invention, the unit control apparatus includes：Product monitoring unit, it is monitored described
The quantity of the product actually manufactured in unit, when according to by the part supply condition monitoring unit monitor it is the multiple
Various parts is quantity determination, the product should being manufactured in the unit in the part of species and the multiple species
It is described when quantity is less than the quantity of the product being monitored by the product monitoring unit, actually being manufactured in the unit
Notification unit is notified to the upper Management Controller.
According to the 3rd invention, in the 1st invention, the unit control apparatus includes：Product monitoring unit, it is monitored described
The quantity of the product actually manufactured in unit, when according to by the part supply condition monitoring unit monitor it is the multiple
Various parts is quantity determination, the product should being manufactured in the unit in the part of species and the multiple species
Quantity is identical with the quantity of the product being monitored by the product monitoring unit, actually being manufactured in the unit, and
When the quantity for the product that should be manufactured in the unit is less than the quantity of the desired product, the notification unit is to institute
Upper Management Controller is stated to be notified.
According to the 4th invention, in any one invention in the 1 to the 3rd, the production system possesses：Rote learning device,
It learns the creation data of the production system, and the rote learning device possesses：Quantity of state observation unit, it observes the production system
Quantity of state；The result of the action obtaining section, it obtains the manufacture result of product described in the production system；Study portion, it is received
Output from the quantity of state observation unit and the output from the result of the action obtaining section, by the creation data and institute
The quantity of state and the manufacture result for stating production system are associated and learnt；And it is intended to determination section, it is with reference to institute
The creation data that rote learning device learns is stated, creation data is exported.
According to the 5th invention, in the 4th invention, the unit control apparatus includes：Product monitoring unit, it is monitored described
The quantity of the product actually manufactured in unit, the quantity of state of the quantity of state observation unit observation is included in following quantity of state
At least one：The quantity of desired product, by the product information monitoring unit monitor the product information, by the part
The quantity of various parts in the part and the multiple species of the multiple species that supply condition monitoring unit is monitored, by described
The multiple machinery that the quantity and the unit of the product for the actual manufacture that product monitoring unit is monitored are included is each
From setting.
According to the 6th invention, in the 4th or the 5th invention, the creation data for being intended to determination section output includes data below
In at least one：The quantity of various parts should be supplied into the multiple species of at least one unit and described
The multiple mechanical setting that at least one unit is included.
According to the 7th invention, in the 4th invention, the rote learning device possesses：Learning model, it learns creation data,
The rote learning device possesses：Error calculation portion, its calculate the manufacture result that described the result of the action obtaining section obtains with it is pre-
Error between the target first set；And learning model update section, it updates the learning model according to the error.
According to the 8th invention, in the 4th invention, the rote learning device has the value for the value for determining creation data
Function, the rote learning device is also equipped with：Return calculating part, its manufacture knot obtained in the result of the action obtaining section
In the case that difference between fruit and target set in advance is small, positive return is given according to the difference, big in the difference
In the case of, negative return is given according to the difference；And cost function update section, it updates the value according to the return
Function.
The detailed description of typical embodiment of the invention shown with reference to the accompanying drawings, makes object of the present invention, spy
Levy and advantage and other purposes, feature and advantage become definitely.
Brief description of the drawings
Fig. 1 is the block diagram of the production system based on the present invention.
Fig. 2 is the figure for the example for representing product information.
Fig. 3 is the flow chart for the action for representing the production system based on the present invention.
Fig. 4 is the figure of one for representing rote learning device.
Fig. 5 is the figure for another example for representing rote learning device.
Fig. 6 is the figure for schematically showing neuron models.
Fig. 7 is the figure for schematically showing 3 layers of neutral net that the neuron shown in Fig. 6 is combined and constituted.
Fig. 8 is the block diagram of production system of the prior art.
Embodiment
Below, embodiments of the present invention are illustrated referring to the drawings.In following accompanying drawing, to identical part mark
Note identical reference marks.It is appropriate in the drawings to change engineer's scale in order to be readily appreciated that.
Fig. 1 is the block diagram of the production system based on the present invention.Production system 10 possesses：Unit 40, it includes at least one,
It is preferred that multiple (being two in illustrated example) machinery R1, R2 and at least one being controlled to mechanical R1, R2 (are typically and machine
Tool identical quantity) machine control unit (numerical control device) RC1, RC2；Unit control apparatus (cell controller) 30,
Consisting of can communicate with each machine control unit RC1, RC2；And controlled as the upper management of production schedule device
Device 20, it can communicate with unit control apparatus 30.These mechanical R1, R2 individually or cooperate by multiple species part
Manufacture product.Machine control unit RC1, RC2 carry out mechanical R1, R2 action control respectively, and will be surveyed in each machinery
Fixed data are sent to unit control apparatus 30.
Unit 40 is multiple mechanical set for implementing predetermined operation.In addition, machinery R1, R2 are, for example, machine
Bed, articulated robot, parallel robots, manufacture machinery, industrial machine etc., it is each it is mechanical each other can with it is identical can not also
Together.Also, the unit 40 ', 40 " of same structure is connected to unit control apparatus 30.
In Fig. 1, sensor S1, S2 are installed on mechanical R1, R2 respectively.These sensors S1, S is to mechanical R1, R2
Speed, acceleration-deceleration, at least one in Acceleration and deceleration time detected.Also, possess the production to having manufactured in unit 40
The sensor S3 that the various quality of product are detected.
In addition, in the present invention, unit 40,40 ', 40 " may be disposed in factory of manufacture product etc., on the other hand, single
First control device 30 and upper Management Controller 20 may be provided in buildings different from factory etc..In this case, unit
Control device 30 and machine control unit RC1, RC2 can be connected via the networks such as in-house network (the first communication unit 41).In addition, on
Position Management Controller 20 may be disposed in office away from factory etc..In this case, upper Management Controller 20 can be through
By the networks such as internet (the second communication unit 42), it is communicatively coupled with unit control apparatus 30.But this is an example, the
As long as a communication unit 41 is communicatively coupled unit control apparatus 30 and machine control unit RC1, RC2, then any form all may be used
With as long as in addition, the second communication unit 42 is communicatively coupled unit control apparatus 30 and upper Management Controller 20 then any shape
Formula can.
Upper Management Controller 20 is, for example, personal computer, as manufacture system 10 the production schedule and to unit
Control device 30 transmit production schedule device and play a role.As shown in figure 1, upper Management Controller 20 includes product information
S0。
Fig. 2 is the figure for the example for representing product information S0.Product information S0 represents manufacture one by the form of chart (map)
The species of part required for individual product and the quantity of all parts.In the example shown in Fig. 2, a product is by three kinds of parts
A~C is constituted.Also, manufacture one using the part C of the components A of NA0 quantity, the part B of NB0 quantity and NC0 quantity
Individual product.
Operator is inputted the quantity N0 of desired product to upper Management Controller 20 using input unit etc..Upper pipe
Quantity N0 of the controller 20 based on the feedback from unit control apparatus 30 and desired product is managed, to control multiple species
Components A~C supply to unit 40,40 ', 40 ".In addition, product information S0 can also include the quantity N0 of desired product.
Unit control apparatus 30 is configured to be controlled unit 40,40 ', 40 ", specifically, can be to Mechanical course
Device RC1, RC2 send various instructions, or obtain from machine control unit RC1, RC2 each mechanical R1, R2 running status
Etc. data.
As shown in figure 1, unit control apparatus 30 includes：Product information monitoring unit 31, it monitors product information S0；Part is supplied
To Stateful Inspection portion 32, it monitors the number of supply various parts into the part and multiple species of multiple species of the grade of unit 40
Amount；And product monitoring unit 33, it monitors the quantity of the product actually manufactured in unit 40,40 ', 40 ".Also, unit control
Device 30 processed is included in the notification unit 34 notified when causing predetermined state as problem to upper Management Controller 20.And
And, unit control apparatus 30 includes rote learning device 50 described later.Rote learning device 50 can also be contained in upper management
In controller 20.In addition, rote learning device 50 can also be connected to unit control apparatus 30 or upper management control externally
Device 20.
Fig. 3 is the flow chart for the action for representing the production system based on the present invention.Below, with reference to these accompanying drawings, to production
The action of system 10 is illustrated.Content shown in Fig. 3 is, when production system 10 is run, to repeat real by predetermined controlling cycle
The content applied.In addition, in following example, for the purpose of succinct, being set to only manufacture product in unit 40.It is noted that also may be used
To carry out substantially same control in unit 40 ', 40 ".
First, in step s 11, the product information monitoring unit 31 of unit control apparatus 30 obtains upper Management Controller 20
Product information S0 and desired product quantity N0.Then, in step s 12, the part supply of unit control apparatus 30
The monitoring parts A of Stateful Inspection portion 32~C supply condition.In other words, part supply condition monitoring unit 32 obtains supply to list
Various parts A~C quantity NA1, NB1, NC1 in the components A~C and multiple species of multiple species of member 40.
Then, in step s 13, judge whether all parts A~C is supplied suitably to unit 40.For each portion
Part A~C, sets the upper limit number that can be suitably processed in unit 40 and lower limit number.In step s 13, judging part A~C
Whether quantity NA1~NC1 is between corresponding upper limit number and lower limit number.
Then, for example it is bigger or smaller than corresponding lower limit number than corresponding upper limit number in the quantity NA1 of components A
In the case of, it is transferred to step S15.In step S15, judgement part A supply amount is superfluous or not enough, and notification unit 34 should
Information is notified to upper Management Controller 20.Same processing is also carried out for miscellaneous part B, C.
As described above, in order to manufacture a product, it is necessary to all components A~C of multiple species.Therefore, be judged as it is many
When the quantity of at least one part in components A~C of individual species is superfluous or deficiency, it is impossible to manufacture product well, notify
Portion 34 is notified to upper Management Controller 20.
In this case, components A~C of upper 20 pairs of surpluses of Management Controller or deficiency quantity, which for example increases, adds deduct
Few predetermined quantity.It therefore, it can make production system 10 effectively act.
In addition, in step s 13, when the quantity NA1~NC1 for being determined as components A~C is in corresponding upper limit number with
When between limit number, it can be determined that for components A~C can be used suitably to manufacture product.Therefore, now, step S14 is transferred to, after
The continuous manufacture for carrying out product.
Then, in step s 16, the quantity N1 for the product that can be manufactured by unit 40 is calculated.Obtained according in step S11
Product information S0 and step S12 in components A~C quantity NA1~NC1 for obtaining be determined to be manufactured by unit 40
Product quantity N1.
Next, in step S17, the product monitoring unit 33 of unit control apparatus 30 is obtained by the actual manufacture of unit 40
The quantity N2 of product.Also, in step S18, judge whether the quantity N1 for the product that can be manufactured by unit 40 is less than reality
Whether the quantity N1 of the quantity N2 of the product of manufacture and the product that can be manufactured by unit 40 is more than the product of actual manufacture
Quantity N2.
When the quantity N1 for the product for being judged to be manufactured by unit 40 in step S18 is less than the product of actual manufacture
During quantity N2, it can be determined that there occurs failure at least one in multiple mechanical R1, R2 in unit 40.Therefore, notification unit
34 notify the situation in step S19 to upper Management Controller 20.Then, upper Management Controller 20 is for example with identical ratio
Example reduces the quantity of the part of multiple species, thus, production system 10 is effectively acted.
In addition, the quantity N1 for the product that can be manufactured by unit 40 is more than the quantity N2 of the product of actual manufacture situation,
It will not actually be caused.Therefore, in step S18, in the case where being judged as described above, notification unit 34 is to upper pipe
Reason controller 20 notifies that exception (step S19) may be there occurs in unit 40.
In addition, quantity N1 and the production of actual manufacture when the product for being judged to be manufactured by unit 40 in step S18
When the quantity N2 of product is equal, it can be determined that not occur exception in unit 40.Desired by now, being obtained in step S20
The quantity N0 of product, judges whether the quantity N1 for the product that can be manufactured by unit 40 is less than desired production in the step s 21
The quantity N0 of product.In addition it is also possible to omit step S20 processing.
It is equal with the quantity N2 of the product of actual manufacture in the quantity N1 for the product that can be manufactured by unit 40, but can
When the quantity N1 of the product manufactured by unit 40 is less than the quantity N0 of desired product, it can be determined that for supply to unit 40
Components A~C quantity is few.Therefore, notification unit 34 notifies the situation to upper Management Controller 20.Then, upper management control
Device 20 for example increases A~C of the part of multiple species quantity with predetermined ratio, thus, production system 10 is effectively acted.
So, unit control apparatus 30 of the invention using product information monitoring unit 31, part supply condition monitoring unit 32,
Product monitoring unit 33, various information are obtained from upper Management Controller 20 and unit 40.Then, unit control apparatus 30 is based on each
It is abnormal to judge whether to occur to plant information, when having abnormal, is notified to upper Management Controller 20.Due in this hair
Exception can be eliminated in bright rapidly, therefore, production system 10 is effectively acted.
In addition, Fig. 4 is the figure of one for representing rote learning device.In the present invention, monitored using from product information
The information in portion 31, part supply condition monitoring unit 32 and product monitoring unit 33 is learnt rote learning device 50.Mechanics
Device 50 is practised to possess quantity of state observation unit 11, the result of the action obtaining section 12, study portion 13 and be intended to determination section 14.
The study portion 13 of rote learning device 50 receives the quantity of state being observed from the quantity of state to production system 10
The output of observation unit 11 and the output (production from the result of the action obtaining section 12 for obtaining the processing result in production system 10
The manufacture result of product), creation data and the quantity of state and manufacture result of production system 10 are associated and learnt.And
And, it is intended that the creation data that determination section 14 learns with reference to study portion 13, to determine creation data and to unit control apparatus 30
Output.
Here, the quantity of state that quantity of state observation unit 11 is observed includes the quantity N0 of desired product, by product information
Product information S0, components A~C of the multiple species monitored by part supply condition monitoring unit 32 and multiple that monitoring unit 31 is monitored
In species quantity NA1~NC1 of various parts, the quantity N2 of the product of the actual manufacture monitored by product monitoring unit 33 and
At least one in multiple mechanical R1, R2 for being included in unit 40 respective setting.In addition, multiple mechanical R1, R2 setting
Responsiveness, acceleration-deceleration, Acceleration and deceleration time including mechanical R1, R2 etc..
In addition, it is intended that the creation data that determination section 14 is exported is included and should be supplied in multiple species of at least one unit 40
In multiple mechanical R1, R2 for being included in the quantity NA2~NC2 and at least one unit 40 of various parts setting at least
One.
Study portion 13 possesses the learning model for learning different creation datas.Also, study portion 13 possesses：Error calculation portion
15, the manufacture result acquired by its calculating action result obtaining section 12, the various quality of quantity, product such as product, with
Error between target set in advance；And learning model update section 16, it is according to error come renewal learning model.
Based on some creation data come during manufacturing product, if being used as the output from the result of the action obtaining section 12
One kind and the product quality that receives exceedes predetermined threshold, then manufacture result of the output hypothesis of error calculation portion 15 in creation data
In generate the result of calculation of predictive error.Also, learning model update section 16 is according to the result of calculation come renewal learning model.
In addition, Fig. 5 is the figure for the other examples for representing rote learning device.Study portion 13 shown in Fig. 5 possesses：Return meter
Calculation portion 18 and according to return come the cost function update section 19 of recovery value function.Also, the rote learning dress shown in Fig. 5
Put 50 data recording sections 17 for not possessing spin-off (label).According to the content of the manufacture result of product, by each production number
Contents different according to this possesses the cost function for the value for determining creation data.
Return difference of the calculating part 18 between the product quality and aimed quality that the result of the action obtaining section 12 is obtained small
In the case of, positive return is given according to the small degree, in the case where difference is big, negative return is given according to the big degree
Report.
In addition, now, preferably during product is manufactured with some creation data, if being taken as from the result of the action
The product quality for obtaining one kind of the output in portion 12 and receiving exceedes predetermined threshold, then returns calculating part 18 and give predetermined bear back
Report, cost function update section 19 is according to predetermined negative return come recovery value function.
Finally, the learning method to rote learning device 50 is described.Rote learning device 50 possesses following function：It is logical
Parsing is crossed to extract useful rule of the input into the data acquisition system of rote learning device 50, specification, Knowledge representation, judge base
Standard etc., exports the judged result, and carry out the study of knowledge.
Include teacher learning, teacherless learning, intensified learning scheduling algorithm in rote learning, also, realize these
On the basis of method, with learning characteristic amount side extraction, being referred to as " Deep Learning (Deep Learning) " of itself
Method.
It is following method to have teacher learning：By largely providing certain input and result (label) to rote learning device 50
The group of data, learns the feature in these data sets, and the model that result is estimated according to input, i.e. its relation are obtained by concluding
Property.There is teacher learning by providing appropriate input data and pair of output data for learning, can relatively easily promote
Study.
Teacherless learning is following method：By only largely providing input data to learning device, study input data is such as
What is distributed, even if not providing corresponding teacher's output data, also can by being compressed, classifying to input data, shaping etc.
Device learnt." content that should be exported " this respect is not being predetermined, from there is teacher learning different.It can be used for extracting
It is present in the construction substantially of data behind.
Intensified learning is following method：In addition to judging or classifying, by learning to behavior, given based on to environment
The interaction of behavior learns appropriate behavior, i.e., to for making the learning method of getable return maximum in the future
Practise.In intensified learning, although learned since the result caused by behavior or the state of endless all-knowingness is not known completely
Practise, but can also be excellent from condition using the state of prior learning as original state by there is teacher learning to be learnt in advance
Beginning place start intensified learning.Intensified learning has the behavior that can evenly select to open up unknown learning region and profit
The feature of the behavior carried out with known learning region, may be further discovered that in past completely ignorant criteria range
It is used as the appropriate working condition of purpose.In addition, because of the output of creation data, the change, i.e. behavior such as temperature of machinery or product
Environment is impacted, it is taken as that meaningful using intensified learning.
Fig. 4 indicates the example of the rote learning device 50 of teacher learning, and Fig. 5 represents the rote learning device of intensified learning
50 example.
First, to being described based on the learning method for having teacher learning.There is provided for study in having teacher learning
Appropriate input data and pair of output data, generation is to input data and answers the letter that corresponding output data is mapped
Number (learning model).
The action for carrying out having the rote learning device of teacher learning has study stage and this 2 stages of forecast period.Carry out
There is the rote learning device of teacher learning in the study stage, if providing comprising the state variable (explanatory variable) as input data
Value and purpose variable as output data value teacher's data, then in the value of input state variable, to output purpose
The situation of the value of variable is learnt, by providing several this teacher's data, is used to export and state variable value pair to construct
The forecast model for the purpose variate-value answered.Also, have the rote learning device of teacher learning in forecast period, it is new when providing
During input data (state variable), according to learning outcome (forecast model constructed), predict and export output data (purpose change
Amount).Here, the data recording section 17 of spin-off (label) can keep the data of the spin-off (label) so far obtained,
And the data of spin-off (label) are supplied to error calculation portion 15.Or, can also by storage card or communication line etc.,
The data of the spin-off (label) of unit control apparatus 30 are supplied to the error calculation portion 15 of the unit control apparatus 30.
As one of the study for the rote learning device for have teacher learning, for example, setting shown in following formula (1)
Forecast model regression equation, and will in learning process each state variable x1, x2, x3... when the value taken substitutes into regression equation,
By adjusting each coefficient a0, a1, a2, a3... value promote study, to obtain purpose variable y value.In addition, learning method
It is not limited thereto, it is different by the algorithm for having teacher learning.
Y=a0+a1x1+a2x2+a3x3+…+anxn
As the algorithm for having teacher learning, it is known to neutral net, least square method, stepwise process (stepwise
) etc. method various methods, are used as the method applied to the present invention, it would however also be possible to employ any to have teacher learning algorithm.In addition,
Because each has teacher learning algorithm to be known, therefore omit the detailed description of each algorithm in this specification.
Then, the learning method based on intensified learning is described.The problem of as intensified learning, sets, by such as lower section
Formula is thought deeply.
The ambient condition of the states of 13 pairs of study portion comprising unit 40 is observed, come determine behavior (creation data
Output).
Environment changes according to certain rule, also, behavior sometimes also can produce change to environment.
When implementing behavior every time, return signal is returned.
Want it is maximized be future return it is total.
Since do not know completely or endless all-knowingness behavior caused by result state learn.
It is used as the exemplary process of intensified learning, it is known that Q learns or TD study.Below, said in the case where Q learns
It is bright, but also it is not limited to Q study.
Q study is finished classes and leave school the value Q (s, method a), in certain state s that commonly use in housing choice behavior a in certain ambient condition s
When, by value Q, (s, a) highest behavior a selections are optimal behavior.But, initially, for state s and behavior a group
Close, in this case it is not apparent that (s, right value a), therefore intelligent body (behavioral agent) select various actions a, pin to value Q under certain state s
Return is provided to behavior a now.Thus, intelligent body selects more preferably behavior, i.e. and the correct value Q of study (s, a).
Also, the result of behavior, it is desirable to make the total maximization of return obtained in the future, therefore with final Q (s, a)=E
[Σ(γt)rt] it is target.Here, E [] represents expected value, t is the moment, and γ is the parameter described later for being referred to as discount rate, rtFor
Moment t return, Σ is the total of moment t.Expected value in the formula is taken when being according to optimal behavior generating state change
Value, it is not known that the value, therefore explore while learning.For example, value Q as being represented by following formula (1) (s, a)
Newer.
That is, cost function update section 21 carrys out recovery value function Q (s using following formula (2)t, at)。
Wherein, stRepresent the ambient condition at moment t, atRepresent the behavior at moment t.Pass through behavior at, state change is st+1。
rt+1Represent the return obtained according to the state change.In addition, the item with max is in state st+1The lower Q for selecting now to know
Q values during value highest behavior a are multiplied by item obtained by γ.Here, γ is the parameter of 0 ＜ γ≤1, referred to as discount rate.In addition, α
For learning coefficient, the scope of 0 ＜ α≤1 is set to.
Above-mentioned formula (2) is expressed as follows method：According to trial atResult and the return r that returnst+1, more new state stUnder
Behavior atEvaluation of estimate Q (st, at).That is, if representing based on return rt+1With the optimal behavior max a under behavior a next state
Evaluation of estimate Q (st+1, max at+1) total behavior a being more than under state s evaluation of estimate Q (st, at), then by Q (st, at) set
To be larger, if on the contrary, evaluation of estimate Q (s less than the behavior a under state st, at), then by Q (st, at) it is set to smaller.In other words
Say, make the value of certain behavior under certain state close to based on the return and the next state of the behavior returned as a result and immediately
Under optimal behavior value.
Here, (s, the technique of expression on computer a) has Q：For all state behaviors to (s, a), using its value as
Behavior memory table and the method kept；And prepare approximate Q (s, the method for function a).In the method for the latter, it can pass through
The methods such as probability gradient descent method adjust the parameter of approximate function, thus, it is possible to realizing above-mentioned formula (2).In addition, as approximate
Function, can use neutral net.
, can be with as described above, as the approximate data of the cost function in the learning algorithm, intensified learning for having teacher learning
Using neutral net, therefore preferably, rote learning device 50 has neutral net.
Fig. 6 is the figure for schematically showing neuron models, and Fig. 7 is to schematically show to enter the neuron shown in Fig. 6
The figure for the three-layer neural network that row is combined and constituted.Neutral net as the neuron shown in simulation drawing 6 model arithmetic unit
And memory etc. is constituted.Output (result) y of neuron output for multiple input x.To each input x (x1~x3) be multiplied by with being somebody's turn to do
Input the corresponding weight w (w of x1~w3), the result y that neuron output is showed by following formula (3).In addition, input x, result y
And weight w is vector.
Wherein, θ is biasing, fkFor activation primitive.
As shown in fig. 7, inputting multiple input x (x from the left side of neutral net1~x3), from right side output result y (y1~
y3).Input x1~x3It is multiplied by after corresponding weight and is input to 3 neuron N11~N13Each neuron.To these inputs
The weight unified presentation of multiplication is w1。
Neuron N11~N13Z is exported respectively11~z13.In the figure 7, by these z11~z13Unified presentation is characterized vector
z1, can be regarded as being extracted vectorial obtained by the characteristic quantity of input vector.This feature vector z1For weight w1With weight w2Between
Characteristic vector.z11~z13It is multiplied by after corresponding weight and is separately input to 2 neuron N21And N22Each neuron.Will
These weight unified presentations being multiplied by characteristic vector are w2.Neuron N21, N22Z is exported respectively21, z22.In the figure 7, by these
z21, z22Unified presentation is characterized vectorial z2.This feature vector z2For weight w2With weight w3Between characteristic vector.z21, z22Multiply
To be separately input to 3 neuron N after corresponding weight31~N33Each neuron.These are multiplied by characteristic vector
Weight unified presentation is w3。
Finally, neuron N31~N33Difference output result y1~result y3.In the action of neutral net, there is mode of learning
With value forecasting pattern, in mode of learning, learn weight w using learning data set, its parameter is used in predictive mode
Judge the behavior of output creation data.Here, on-line study and batch study can be carried out, wherein, on-line study is instant learning
Data obtained from the output of creation data are actually carried out under predictive mode, and are reflected to next behavior；It is to use to criticize study
The data group collected in advance carries out unified study, carries out detection pattern with the parameter always later.Whenever a certain degree of data
Mode of learning can also be inserted when overstocked.
Furthermore it is possible to learn weight w by error back propagation method (Back propagation)1~w3.In addition, error
Information enters cocurrent to the left from right side.Error back propagation method is following method：For each neuron, adjustment (study) is each
Individual weight is so that the difference of the output y and really output y (teacher) when have input input x diminish.
It is 2 by intermediate layer although the intermediate layer (hidden layer) of Fig. 7 neutral net is one layer, but it is also possible to be set to more than 2 layers
Situation more than layer is referred to as Deep Learning.
More than, to there is the learning method of teacher learning and intensified learning to carry out simple narration, but applied to this hair
Bright learning by rote is not limited to these methods, can also be using the method that can be used in rote learning device 10
That is various methods such as " having teacher learning ", " teacherless learning ", " partly having teacher learning " and " intensified learning ".
Such rote learning device 50 is based on from product information monitoring unit 31, part supply condition monitoring unit 32 and production
The information of product monitoring unit 33 is learnt, thus it is speculated that components A~C of multiple species of each product of manufacture quantity.According to
Components A~C guess value calculates the quantity N1 for the product that can be manufactured by unit 40, by the product of itself and actual manufacture
Quantity N2 compares.Also, with the above likewise it is possible to speculate that some in mechanical R1, R2 there occurs failure etc..
Also, all parts A~C of 50 pairs of multiple species from part supply condition monitoring unit 32 of rote learning device
Quantity NA1, NB1, NC1 and the quantity N2 time passage of product from product monitoring unit 33 learnt, to speculate
The situation of unit 40.Also, when use up supply to unit 40 components A~C when, thus it is speculated that for part supply stagnate.In addition, nothing
It is whether sufficient by the quantity of the part supplied, when the quantity N2 of the product actually manufactured is smaller, mechanical R1, R2 can be carried out
In some there occurs the supposition of failure etc..
Rote learning device 50 is excellent in terms of real-time, and supervises scope for locality, therefore can improve above-mentioned
Abnormal accuracy of detection.
In addition, the production system 10 of above-mentioned embodiment to 1 production system 10 as shown in figure 1, possess 1 rote learning
Device 50.But, in the present invention, production system 10 and the respective quantity of rote learning device 50 are not limited to 1.It is excellent
, there are multiple production systems 10, the multiple rote learning devices 50 set respectively by production system 10 are via communication media, phase in choosing
Mutually shared or exchange data., can be shorter by sharing the data for including learning outcome acquired by each production system 10
High-precision results of learning are obtained in time, more suitably creation data can be exported.
Also, rote learning device 50 may reside in production system 10, it may reside in outside production system 10.Separately
Outside, single rote learning device 50 can also be shared in multiple production systems 10 via communication media.Also, mechanics
Device 50 is practised to be also present on Cloud Server.
As a result, can not only share results of learning, data can also be managed concentratedly, and using on a large scale
High-performance processor is learnt.Therefore, pace of learning, the precision of study are improved, and can export more suitably creation data.
In addition, can also shorten the time determined required for the creation data to be exported.Although also may be used in these rote learning devices 50
To use general computer or processor, if but using GPGPU (General-Purpose computing on
Graphics Processing Units, general-purpose computations graphics processor) or extensive PC clusters etc., then can be more at high speed
Handled.
In the 1st invention, when multiple species all parts quantity not in respective preset range when, it can be determined that
To supply at least one superfluous or deficiency into the part of multiple species of unit.Therefore, notified to upper Management Controller
The situation, upper Management Controller suitably changes the quantity of superfluous or not enough part, thus, makes the effective earthquake of production system
Make.
In the 2nd invention, can be determined according to the quantity of the part of multiple species of supply to unit can be by unit style
The quantity for the product made.Also, the actual system monitored by product monitoring unit can be less than by the quantity of the product of unit making
During the quantity for the product made, it can be determined that there occurs failure at least one in multiple machineries in unit.Therefore, to upper
Management Controller notifies the situation, and upper Management Controller is reduced the quantity of the part of multiple species with same ratio, thus, made
Production system is effectively acted.
In the 3rd invention, that is, allow to the quantity and the actual system that is monitored by product monitoring unit by the product of unit making
The quantity for the product made is identical, can when that can be less than the quantity of desired product by the quantity of the product of unit making
It is judged as supply to the lazy weight of the part of multiple species of unit.Therefore, the situation is notified to upper Management Controller, on
Position Management Controller increases the quantity of the part of multiple species, thus, production system is effectively acted.
In the invention of 4 to the 8th, the abnormal accuracy of detection in production system can be improved.
The present invention is illustrated using typical embodiment, still, if those skilled in the art, this is not being departed from
Under conditions of the scope of invention, above-mentioned change and various other changes, omission, addition can be carried out.
Claims (8)
1. a kind of production system, it is characterised in that possess：
At least one unit, its multiple machinery for including manufacturing product and the multiple Mechanical courses being controlled to the plurality of machinery
Device；
Unit control apparatus, it communicably is connected to control the unit with least one unit；And
Upper Management Controller, it is communicably connected with the unit control apparatus, and including product information,
The product information includes being used to manufacture various in the part of multiple species of a product and the multiple species
The quantity of part,
The unit control apparatus possesses：
Product information monitoring unit, it monitors the product information；
Part supply condition monitoring unit, it monitors supply to the part of the multiple species of at least one unit and described
The quantity of various parts in multiple species；And
Notification unit, the quantity of its various parts in the multiple species monitored by the part supply condition monitoring unit does not exist
In the case of in the preset range determined according to various parts in the multiple species, led to the upper Management Controller
Know.
2. production system according to claim 1, it is characterised in that
The unit control apparatus includes：Product monitoring unit, it monitors the number of the product actually manufactured in the unit
Amount,
When in the part and the multiple species according to the multiple species monitored by the part supply condition monitoring unit
The quantity of the product that the quantity of various parts is determined, should being manufactured in the unit is less than is supervised by the product monitoring unit
Depending on to, the quantity of the product that actually manufactures in the unit when, the notification unit is to the upper Management Controller
Notified.
3. production system according to claim 1, it is characterised in that
The unit control apparatus includes：Product monitoring unit, it monitors the number of the product actually manufactured in the unit
Amount,
When in the part and the multiple species according to the multiple species monitored by the part supply condition monitoring unit
The quantity of the product that the quantity of various parts is determined, should being manufactured in the unit by the product monitoring unit with being monitored
To, the quantity of the product that actually manufactures in the unit it is identical, and the production that should be manufactured in the unit
When the quantity of product is less than the quantity of the desired product, the notification unit is notified to the upper Management Controller.
4. production system according to any one of claim 1 to 3, it is characterised in that
The production system possesses：Rote learning device, it learns the creation data of the production system,
The rote learning device possesses：
Quantity of state observation unit, it observes the quantity of state of the production system；
The result of the action obtaining section, it obtains the manufacture result of product described in the production system；
Study portion, it receives the output from the quantity of state observation unit and the output from the result of the action obtaining section,
The creation data is associated and learnt with the quantity of state and the manufacture result of the production system；And
It is intended to determination section, its creation data learnt with reference to the rote learning device exports creation data.
5. production system according to claim 4, it is characterised in that
The unit control apparatus includes：Product monitoring unit, it monitors the number of the product actually manufactured in the unit
Amount,
The quantity of state of the quantity of state observation unit observation includes at least one in following quantity of state：The number of desired product
Amount, the product information monitored by the product information monitoring unit, the institute monitored by the part supply condition monitoring unit
State the quantity of various parts in the part and the multiple species of multiple species, the actual system monitored by the product monitoring unit
The multiple respective setting of machinery that the quantity and the unit for the product made are included.
6. the production system according to claim 4 or 5, it is characterised in that
The creation data for being intended to determination section output includes at least one in data below：Should supply to it is described at least one
What the quantity of various parts and at least one described unit were included in the multiple species of unit is the multiple mechanical
Setting.
7. rote learning device according to claim 4, it is characterised in that
The rote learning device possesses：Learning model, it learns creation data,
The rote learning device possesses：
Error calculation portion, it is calculated between the manufacture result and target set in advance of the result of the action obtaining section acquirement
Error；And
Learning model update section, it updates the learning model according to the error.
8. rote learning device according to claim 4, it is characterised in that
The rote learning device has the cost function for the value for determining creation data,
The rote learning device is also equipped with：
Calculating part is returned, it is between the manufacture result and target set in advance that the result of the action obtaining section is obtained
In the case that difference is small, positive return is given according to the difference, in the case where the difference is big, is given according to the difference negative
Return；And
Cost function update section, it updates the cost function according to the return.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2016082284A JP2017191567A (en) | 2016-04-15 | 2016-04-15 | Production system for implementing production plan |
| JP2016-082284 | 2016-04-15 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107301489A true CN107301489A (en) | 2017-10-27 |
Family
ID=59981003
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710243910.6A Pending CN107301489A (en) | 2016-04-15 | 2017-04-14 | Implement the production system of the production schedule |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US20170300041A1 (en) |
| JP (1) | JP2017191567A (en) |
| CN (1) | CN107301489A (en) |
| DE (1) | DE102017003427A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115702401A (en) * | 2020-07-17 | 2023-02-14 | 三菱电机株式会社 | Scheduler system, scheduler management device, and machine learning device |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2019160176A (en) * | 2018-03-16 | 2019-09-19 | ファナック株式会社 | Component supply amount estimating apparatus, and machine learning apparatus |
| JP7236127B2 (en) * | 2018-07-20 | 2023-03-09 | 中日本炉工業株式会社 | Processing equipment operation management system |
| JP6663064B1 (en) * | 2019-05-17 | 2020-03-11 | ｓｇｌａｂ株式会社 | Order management device, order management method and order management program |
| JP7256703B2 (en) * | 2019-06-21 | 2023-04-12 | ファナック株式会社 | Controllers, control systems, and machine learning devices |
| JP2021116783A (en) * | 2020-01-29 | 2021-08-10 | トヨタ自動車株式会社 | Vehicular control device and vehicular control system |
| JP2021116781A (en) * | 2020-01-29 | 2021-08-10 | トヨタ自動車株式会社 | Vehicle control method, vehicular control device and server |
| JP7359011B2 (en) * | 2020-02-05 | 2023-10-11 | トヨタ自動車株式会社 | Internal combustion engine control device |
| JP6869439B1 (en) * | 2020-02-25 | 2021-05-12 | 三菱電機株式会社 | Monitoring system and monitoring method |
| EP3928267A4 (en) * | 2020-03-31 | 2022-11-16 | ATS Automation Tooling Systems Inc. | Systems and methods for modeling a manufacturing assembly line |
| CN118195284A (en) * | 2024-05-17 | 2024-06-14 | 深圳星际鑫航科技有限公司 | Intelligent operation and maintenance visual monitoring method and system based on big data |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH04343648A (en) * | 1991-05-20 | 1992-11-30 | Yamatake Honeywell Co Ltd | Many-variety small-quantity production system |
| JP3243745B2 (en) * | 1991-08-26 | 2002-01-07 | 日本電信電話株式会社 | Device control computer and device control method |
| JPH05104395A (en) * | 1991-10-17 | 1993-04-27 | Hokuriku Nippon Denki Software Kk | Product processing condition setting device with learning control function |
| JP3981427B2 (en) * | 1995-08-09 | 2007-09-26 | 富士フイルム株式会社 | Production management method and apparatus |
| TWI281603B (en) * | 2003-11-13 | 2007-05-21 | Amada Co Ltd | Sheet metal factory processing system, processing schedule management method and recording media |
| JP4449448B2 (en) * | 2003-12-24 | 2010-04-14 | パナソニック株式会社 | Component supply management apparatus and component supply management method |
| JP2005339058A (en) * | 2004-05-25 | 2005-12-08 | Masaru Nakasaki | Cell production work management system, cell production work management method, cell production work management program, and recording medium to which the program is recorded |
| JP4775966B2 (en) * | 2007-06-15 | 2011-09-21 | オムロン株式会社 | Assembly support system in a heterogeneous simultaneous production line |
| JP5018641B2 (en) * | 2008-05-26 | 2012-09-05 | オムロン株式会社 | Production management system and production management method |
| JP2011028654A (en) * | 2009-07-28 | 2011-02-10 | Panasonic Electric Works Co Ltd | Production facility to be operated in collaboration with person |
| JP2013016087A (en) | 2011-07-05 | 2013-01-24 | Panasonic Corp | Parts management device and parts management method |
| WO2014148564A1 (en) * | 2013-03-19 | 2014-09-25 | 株式会社イシダ | Quantitative weighing system and quantitative weighing method |
-
2016
- 2016-04-15 JP JP2016082284A patent/JP2017191567A/en active Pending
-
2017
- 2017-04-07 DE DE102017003427.3A patent/DE102017003427A1/en not_active Withdrawn
- 2017-04-12 US US15/485,308 patent/US20170300041A1/en not_active Abandoned
- 2017-04-14 CN CN201710243910.6A patent/CN107301489A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115702401A (en) * | 2020-07-17 | 2023-02-14 | 三菱电机株式会社 | Scheduler system, scheduler management device, and machine learning device |
| CN115702401B (en) * | 2020-07-17 | 2023-07-11 | 三菱电机株式会社 | Scheduler system, scheduler management device, and machine learning device |
Also Published As
| Publication number | Publication date |
|---|---|
| DE102017003427A1 (en) | 2017-10-19 |
| JP2017191567A (en) | 2017-10-19 |
| US20170300041A1 (en) | 2017-10-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107301489A (en) | Implement the production system of the production schedule | |
| CN106557069B (en) | Rote learning apparatus and method and the lathe with the rote learning device | |
| US9958861B2 (en) | Production control system and integrated production control system | |
| US10782664B2 (en) | Production system that sets determination value of variable relating to abnormality of product | |
| US20180003588A1 (en) | Machine learning device which learns estimated lifetime of bearing, lifetime estimation device, and machine learning method | |
| JP6438450B2 (en) | Machine learning apparatus, robot system, and machine learning method for learning processing sequence of laser processing robot | |
| Reynoso-Meza et al. | Controller tuning using evolutionary multi-objective optimisation: current trends and applications | |
| JP6174669B2 (en) | Cell control system, production system, control method and control program for controlling manufacturing cell having a plurality of manufacturing machines | |
| JP6386523B2 (en) | Machine learning device, life prediction device, numerical control device, production system, and machine learning method for predicting life of NAND flash memory | |
| DE102016015017B4 (en) | Control device with learning function for detecting a cause of noise | |
| CN106560751B (en) | Machine learning device, machine learning method and the lathe for possessing machine learning device | |
| US10949740B2 (en) | Machine learning device, numerical controller, machine tool system, manufacturing system, and machine learning method for learning display of operation menu | |
| CN106826812A (en) | Rote learning device and learning by rote, laminated cores manufacture device and system | |
| CN110303492A (en) | Control device, machine learning device and system | |
| JP6333868B2 (en) | Cell control device and production system for managing the operating status of a plurality of manufacturing machines in a manufacturing cell | |
| CN110023850A (en) | Method and control device for control technology system | |
| CN106814606B (en) | Rote learning device, motor control system and learning by rote | |
| CN108723889A (en) | Acceleration/deceleration control device | |
| CN106557816A (en) | Possesses the production equipment of rote learning device and assembling exerciser | |
| El-Bouri et al. | A neural network for dispatching rule selection in a job shop | |
| CN113614743A (en) | Method and apparatus for operating a robot | |
| CN108687766A (en) | Control device, machine learning device and the machine learning method of robot | |
| CN110340884A (en) | Measure action parameter adjustment device, machine learning device and system | |
| Meiners et al. | Application of machine learning for product batch oriented control of production processes | |
| Alvi et al. | Efficient use of hybrid adaptive neuro-fuzzy inference system combined with nonlinear dimension reduction method in production processes |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171027 |